CD36 deficiency attenuates experimental mycobacterial infection

Abstract

Background: Members of the CD36 scavenger receptor family have been implicated as sensors of microbial products that mediate phagocytosis and inflammation in response to a broad range of pathogens. We investigated the role of CD36 in host response to mycobacterial infection.

Methods: Experimental Mycobacterium bovis Bacillus Calmette-Guérin (BCG) infection in Cd36+/+ and Cd36-/- mice, and in vitro co-cultivation of M. tuberculosis, BCG and M. marinum with Cd36+/+ and Cd36-/-murine macrophages.

Results: Using an in vivo model of BCG infection in Cd36+/+ and Cd36-/- mice, we found that mycobacterial burden in liver and spleen is reduced (83% lower peak splenic colony forming units, p < 0.001), as well as the density of granulomas, and circulating tumor necrosis factor (TNF) levels in Cd36-/- animals. Intracellular growth of all three mycobacterial species was reduced in Cd36-/- relative to wild type Cd36+/+ macrophages in vitro. This difference was not attributable to alterations in mycobacterial uptake, macrophage viability, rate of macrophage apoptosis, production of reactive oxygen and/or nitrogen species, TNF or interleukin-10. Using an in vitro model designed to recapitulate cellular events implicated in mycobacterial infection and dissemination in vivo (i.e., phagocytosis of apoptotic macrophages containing mycobacteria), we demonstrated reduced recovery of viable mycobacteria within Cd36-/- macrophages.

Conclusions: Together, these data indicate that CD36 deficiency confers resistance to mycobacterial infection. This observation is best explained by reduced intracellular survival of mycobacteria in the Cd36-/- macrophage and a role for CD36 in the cellular events involved in granuloma formation that promote early bacterial expansion and dissemination.

Figure 1

Cd36^-/-mice have lower bacillary burden relative to Cd36^+/+ controls after challenge with M. bovis BCG. A. Intraperitoneal infection (day 0) with BCG (1.5×10⁷ organisms) resulted in a rise in mycobacterial counts in the spleens of Cd36^+/+(wild type) mice to a maximum at day 14 with subsequent decline. The mycobacterial counts in Cd36^-/- mice (white box, dashed line) were lower overall (p < 0.001), and at specific time points (day 7, *p < 0.01; day 14, *p < 0.001). B and C. Differences between genotypes were less pronounced in the liver (B) and lung (C). Results are displayed as mean ± SEM, with 8 mice per group at each time point, representing two pooled independent experiments.

Figure 2

Histopathological sections of organs of Cd36^+/+ (upper row) and Cd36^-/- (lower row) mice 14 days after IP infection with M. bovis BCG. A and B. Liver sections (H&E stain, 10× magnification) demonstrate more numerous granulomas in Cd36^+/+ mice (arrowheads, A) compared to Cd36^-/- mice (arrowhead, B). C and D. Liver sections (H&E stain, 100× magnification) demonstrate similar microarchitecture of individual granulomas. E and F. Splenic sections (Ziehl-Neelsen stain, 100× magnification) show multiple acid fast bacilli (AFB) in a single field for Cd36^+/+ mice (E), and no visible AFB in Cd36^-/- mice (F). G. Liver granuloma cross-sectional density at day 14 post-infection was lower in Cd36^-/- mice (*p = 0.0038). All granulomas in each histopathological section were counted, with observer blinding, and normalized to the liver cross-sectional area. Data represent mean ± SEM, 4 mice per group.

Figure 3

Co-cultivation of mycobacteria with Cd36^-/- murine macrophages. In vitro, mycobacterial infection of thioglycolate-elicited peritoneal macrophages resulted in lower mycobacterial loads in cultures of Cd36^-/- macrophages (white box, dashed line) compared with wild type controls (black box, solid line). A. Infection with M. tuberculosis (MOI = 10:1) produced a progressive rise in mycobacterial counts over 7 days, with significant difference (p < 0.0001) between groups. B. Similar results were seen after infection with the rapidly growing M. marinum (MOI = 1:1, p < 0.001). Data are shown as mean +/- 95% CI with 4 to 6 replicates at each time.

Figure 4

No difference in uptake of mycobacteria between Cd36^+/+ and Cd36^-/- macrophages. A to F. M. marinum was biotinylated and incubated with Cd36^+/+ (top row) and Cd36^-/- (bottom row) macrophages for 3 hours to allow phagocytosis. Extracellular M. marinum was labeled using streptavidin-conjugated tetramethylrhodamine (TMR) and appears red. Macrophages were fixed and permeabilized and a second streptavidin-conjugated fluorophore (AlexFluor488(c)) was used to label intracellular M. marinum (appears green). A and B. Intracellular (green) M. marinum is readily visible within Cd36^-/- macrophages, with uptake qualitatively equivalent to Cd36^+/+ macrophages. C and D. Control conditions using cytochalasin D (10 μM) to inhibit phagocytosis, demonstrating decreased intracellular (green) mycobacteria in both Cd36^+/+ and Cd36^-/- macrophages. E and F. Control conditions using unbiotinylated mycobacteria, demonstrating the specificity of fluorescent labeling for mycobacteria. G and H. Image analysis demonstrates similar quantities of intracellular (G, green) and extra cellular (H, red) M. marinum (p = 0.29) in Cd36^+/+ and Cd36^-/- macrophages. I and J. Flow cytometry showed no significant difference in the fluorescence intensity associated with intracellular (green) or extracellular (red) M. marinum between Cd36^+/+ (solid line) and Cd36^-/- (dashed line). Shown for comparison are negative control conditions (unbiotinylated M. marinum, no fill color). K and L. In vitro, Cd36^+/+ and Cd36^-/- macrophages showed similar uptake of three mycobacterial species, M. tuberculosis (MOI = 10:1), M. bovis BCG (MOI = 10:1) and M. marinum (MOI = 1:1) (p > 0.05 for all species). Similar uptake was observed at different multiplicities of infection with M. tuberculosis (p > 0.05 for each MOI).

Figure 5

Electron micrographs demonstrating similar numbers of internalized M. marinum bacilli after 6 hours co-incubation with Cd36^+/+ and Cd36^-/- macrophages. A and B. Full cell view demonstrating electron-dense bacilli (arrowheads) within macrophages. C and D. In both cell types, bacilli are contained within a membrane- (black arrowhead) bound phagolysosome (white arrowhead, electron-luscent phagolysosome lumen). E. Enumeration of internalized bacilli demonstrated no significant difference in uptake of M. marinum between Cd36^+/+ and Cd36^-/- macrophages (p = 0.19). Bacilli were counted in a total of 293 macrophages from 3 independent experiments, taking a minimum of 30 random EM fields for each genotype in each experiment.

Figure 6

Production of reactive nitrogen and reactive oxygen intermediates is similar in Cd36^+/+ and Cd36^-/- macrophages. A. Nitric oxide production was measured using the Griess reaction in culture supernatant. In response to stimulation with BCG, nitric oxide production is TLR2 (*p < 0.0001) and IRAK-4 (*p < 0.0001) dependent, but independent of TLR4 (p = 0.16) and CD36 (p = 0.83). B. Similar nitric oxide production by Cd36^+/+ and Cd36^-/- macrophages in response to live and heat-killed BCG and M. marinum, as well as specific toll-like receptor ligands (Pam₃CSK₄ (PAM), a TLR2 ligand, and lipopolysaccharide (LPS); p > 0.05 for all comparisons Cd36^+/+ vs Cd36^-/-). C and D. Similar kinetics of nitric oxide production over 36 hours in response to both BCG (C) and M. marinum (D). E and F. Similar oxidative burst in response to phorbol myristate acetate (PMA), and no significant response to BCG above baseline among Cd36^+/+ and Cd36^-/- macrophages.

Figure 7

Alterations in TNF and IL-10 production do not explain relative resistance of Cd36^-/- mice to mycobacterial infection. A. Serum levels of TNF in Cd36^+/+ and Cd36^-/- mice rose after intraperitoneal infection (day 0) with M. bovis BCG, reaching a maximum after 14-21 days, then decreased. This trend paralleled the mycobacterial counts in organs of infected mice. TNF levels were higher overall (p < 0.001) in Cd36^+/+ mice (solid line) compared to Cd36^-/- mice (dashed line). Results are displayed as mean ± SEM, with each point representing 4 replicate mice. Data from one representative experiment of two are shown. B. Significant correlation (ρ = 0.596, p < 0.001) between splenic mycobacterial counts and serum level of TNF. C. Time course of TNF production by Cd36^+/+ and Cd36^-/- thioglycolate-elicited peritoneal macrophages co-incubated with M. marinum in vitro shows no difference between genotypes (p = 0.33). In contrast, TNF production by Tlr2^-/- (p < 0.0001) and Irak4^-/- (p < 0.0001) macrophages is markedly deficient. D. Cd36^-/- macrophages are not deficient in IL-10 production, and produce higher levels of IL-10 at some time points (* p = 0.040) following infection with M. marinum. E and F. TNF production in response to M. marinum (E) and BCG (F) is similar (p = 0.54 for M. marinum, p = 0.96 for BCG) between Cd36^+/+ and Cd36^-/- macrophages over a range of multiplicities of infection. G and H. IL-10 production in response to M. marinum (G) and BCG (H) was dose-dependent (p < 0.0001 for both) and was higher in Cd36^-/- macrophages (p < 0.0001 for M. marinum, p = 0.0003 for BCG). Cytokine levels were assayed in supernatant of macrophages in 96-well plates (200,000 adherent cells and 250 μl media per well) after 24 hours incubation. These findings were confirmed using bone-marrow derived macrophages (data not shown).

Figure 8

Recovery of viable intracellular mycobacteria is reduced in Cd36^-/- macrophages in an in vitro model recapitulating cellular events in early granuloma formation. Primary "prey" macrophages were incubated with M. marinum and washed to remove extracellular bacteria. Apoptosis was induced by overnight serum starvation, and was evidenced by increased caspase-3/7 activity (data not shown). Control macrophages (no apoptosis) were incubated in media containing 10% fetal bovine serum. Apoptotic primary prey macrophages containing M. marinum were gently scraped and co-incubated with uninfected secondary "predator" macrophages to allow phagocytosis of primary prey macrophages. After washing to remove extracellular apoptotic bodies and bacteria, cells were scraped and plated on mycobacteria culture media for CFU counts. The recovery of viable M. marinum was reduced in systems using Cd36^-/- (white bar) relative to Cd36^+/+ (black bar) macrophages (*p = 0.0026), suggesting a defect in the uptake of mycobacteria-laden apoptotic macrophages.